Nonlinear trend in streamflow and its response to climate change under complex ecohydrological patterns in the Yellow River Basin, China

Natural streamflow (Q) is governed by basin scale interactions between climate, soil, and vegetation. Functional change in the way Q operates is an indicator of climatic and ecohydrological process change. The aim of this study was to explore nonlinear trends in natural Q and its response to changes in precipitation (P) and air temperature (Ta) in China's Yellow River Basin. Temporal trends in natural Q were detected by means of the nonparametric Mann–Kendall model while nonlinear relationships between natural Q, P, and Ta were investigated using the wavelet decomposition and reconstruction method as well as applying multiple regression models. Results show that: (i) decreasing trends have been detected in natural Q, prompted by an abrupt annual scale change that occurred in 1988, and variations of this change in natural Q were evident in the months of October and November and on an annual scale; (ii) natural Q, P, and Ta exhibited nonlinear tendencies for the four different temporal scale treatments applied (2, 4, 8, and 16 years) while decreasing trends were detected for natural Q and P and increasing trends were detected for Ta between 1961 and 2008 on both monthly and annual scales; (iii) annually, natural Q exhibited a significant, positive correlation with P and a negative correlation with Ta in all four temporal scales (p < 0.001). Similar monthly patterns were also observed in October and November. These nonlinear trends that exist between natural Q, P, and Ta will help us to understand the response of hydrological processes to the combined effects of climate change and alterations in ecosystem function as well as to establish applicable ecological regulations and water management practices on a basin scale.

► Analyzed combined effects of climate changes and ecosystem alterations on streamflow.
► Showed the nonlinear tendency of streamflow with Mann–Kendall model.
► Assessed changes in streamflow using multiple regressions model.